January 09, 2018

If AI was allowed to have a crack at the weights used by current risk weighted capital requirements for banks, the regulators would surely have a lot of explaining to do.

Sir, John Thornhill writes that he saw an artificial intelligence program crack in 12 minutes and 50 seconds the mindbendingly complex Enigma code used by the Germans during the second world war” “Competent computers still cannot comprehend” January 9.

I wish AI would also be asked to suggest some weights for the risk weighted capital requirements for banks.

For instance in Basel II the standardized risk weight assigned to something rated AAA, and therefore perceived as very safe, something to which banks could build up dangerous exposures, is 20%; while the risk weight for something rated below BB-, and therefore perceived to be very risky, and therefore banker won’t touch it with a ten feet pole, is 150%.

I would love to see for instance Mario Draghi’s, Mark Carney’s, and Stefan Ingves’ faces, if artificial intelligence, smilingly, came up with weights indicating a quite inverse relation between perceived risks and real dangers to a banking system.